Fuzzy Partition Distance Based Attribute Reduction In Decision Tables

In recent years, researchers have proposed fuzzy rough set based attribute reduction methods direct on original decision tables to improve the accuracy of the classification model. Most of the previously proposed methods are filter methods, which means that the classification accuracy is evaluated after finding reduct. Therefore, the obtained reduct is not optimal both in terms of number of attributes and classification accuracy. In this paper, we propose a fuzzy partitioning distance and a fuzzy partitioning distance based algorithm to find approximate reduct according to filter-wrapper approach. Experimental results on some data sets show that the classification accuracy on reduct of proposed algorithm is more efficient than that of traditional filter algorithms. Furthermore, by using distance measurements, the execution time of the proposed algorithm is more efficient than the execution time of entropy based filter-wrapper algorithms.